基于预测的Web缓存替换策略

Web replacement policy based on prediction

为提高Web缓存性能,在缓存替换算法的基础上加入预测机制,提出了基于预测的Web替换策略P-Re。预测算法采用PPM上下文模型,当缓存空间不够用来存放新的对象时,P-Re选择键值较小且未被预测到的对象进行替换。实验表明,基于预测的W eb缓存替换算法P-Re相对于传统替换算法而言具有较高的命中率和字节命中率。

To improve the performance of Web Cache, a new cache replacement policy named P-Re by importing a prediction mechanism was proposed. Guided by Prediction by Partial Match （PPM） model, P-Re replaced those non-predicted requests with smaller keys when the cache had not enough space to accommodate new requests. The simulation shows that the P-Re replacement algorithm can achieve relatively higher Hit Rate （HR） and Byte Hit Rate （BHR） than the cache replacement policies that have no prediction mechanism.